includes/Export.php

   1 <?php
   2 # Copyright (C) 2003, 2005 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19 /**
  20  *
  21  * @package MediaWiki
  22  * @subpackage SpecialPage
  23  */
  24
  25 /** */
  26 require_once( 'Revision.php' );
  27
  28 define( 'MW_EXPORT_FULL',     0 );
  29 define( 'MW_EXPORT_CURRENT',  1 );
  30
  31 define( 'MW_EXPORT_BUFFER',   0 );
  32 define( 'MW_EXPORT_STREAM',   1 );
  33
  34
  35 /**
  36  * @package MediaWiki
  37  * @subpackage SpecialPage
  38  */
  39 class WikiExporter {
  40         /**
  41          * If using MW_EXPORT_STREAM to stream a large amount of data,
  42          * provide a database connection which is not managed by
  43          * LoadBalancer to read from: some history blob types will
  44          * make additional queries to pull source data while the
  45          * main query is still running.
  46          *
  47          * @param Database $db
  48          * @param int $history one of MW_EXPORT_FULL or MW_EXPORT_CURRENT
  49          * @param int $buffer one of MW_EXPORT_BUFFER or MW_EXPORT_STREAM
  50          */
  51         function WikiExporter( &$db, $history = MW_EXPORT_CURRENT,
  52                         $buffer = MW_EXPORT_BUFFER ) {
  53                 $this->db =& $db;
  54                 $this->history = $history;
  55                 $this->buffer  = $buffer;
  56                 $this->writer  = new XmlDumpWriter();
  57                 $this->sink    = new DumpOutput();
  58         }
  59
  60         /**
  61          * Set the DumpOutput or DumpFilter object which will receive
  62          * various row objects and XML output for filtering. Filters
  63          * can be chained or used as callbacks.
  64          *
  65          * @param mixed $callback
  66          */
  67         function setOutputSink( &$sink ) {
  68                 $this->sink =& $sink;
  69         }
  70
  71         function openStream() {
  72                 $output = $this->writer->openStream();
  73                 $this->sink->writeOpenStream( $output );
  74         }
  75
  76         function closeStream() {
  77                 $output = $this->writer->closeStream();
  78                 $this->sink->writeCloseStream( $output );
  79         }
  80
  81         /**
  82          * Dumps a series of page and revision records for all pages
  83          * in the database, either including complete history or only
  84          * the most recent version.
  85          */
  86         function allPages() {
  87                 return $this->dumpFrom( '' );
  88         }
  89
  90         /**
  91          * Dumps a series of page and revision records for those pages
  92          * in the database falling within the page_id range given.
  93          * @param int $start Inclusive lower limit (this id is included)
  94          * @param int $end   Exclusive upper limit (this id is not included)
  95          *                   If 0, no upper limit.
  96          */
  97         function pagesByRange( $start, $end ) {
  98                 $condition = 'page_id >= ' . intval( $start );
  99                 if( $end ) {
 100                         $condition .= ' AND page_id < ' . intval( $end );
 101                 }
 102                 return $this->dumpFrom( $condition );
 103         }
 104
 105         /**
 106          * @param Title $title
 107          */
 108         function pageByTitle( $title ) {
 109                 return $this->dumpFrom(
 110                         'page_namespace=' . $title->getNamespace() .
 111                         ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) );
 112         }
 113
 114         function pageByName( $name ) {
 115                 $title = Title::newFromText( $name );
 116                 if( is_null( $title ) ) {
 117                         return new WikiError( "Can't export invalid title" );
 118                 } else {
 119                         return $this->pageByTitle( $title );
 120                 }
 121         }
 122
 123         function pagesByName( $names ) {
 124                 foreach( $names as $name ) {
 125                         $this->pageByName( $name );
 126                 }
 127         }
 128
 129
 130         // -------------------- private implementation below --------------------
 131
 132         function dumpFrom( $cond = '' ) {
 133                 $fname = 'WikiExporter::dumpFrom';
 134                 wfProfileIn( $fname );
 135
 136                 $page     = $this->db->tableName( 'page' );
 137                 $revision = $this->db->tableName( 'revision' );
 138                 $text     = $this->db->tableName( 'text' );
 139
 140                 if( $this->history == MW_EXPORT_FULL ) {
 141                         $join = 'page_id=rev_page';
 142                 } elseif( $this->history == MW_EXPORT_CURRENT ) {
 143                         $join = 'page_id=rev_page AND page_latest=rev_id';
 144                 } else {
 145                         wfProfileOut( $fname );
 146                         return new WikiError( "$fname given invalid history dump type." );
 147                 }
 148                 $where = ( $cond == '' ) ? '' : "$cond AND";
 149
 150                 if( $this->buffer == MW_EXPORT_STREAM ) {
 151                         $prev = $this->db->bufferResults( false );
 152                 }
 153                 if( $cond == '' ) {
 154                         // Optimization hack for full-database dump
 155                         $pageindex = 'FORCE INDEX (PRIMARY)';
 156                         $revindex = 'FORCE INDEX(page_timestamp)';
 157                 } else {
 158                         $pageindex = '';
 159                         $revindex = '';
 160                 }
 161                 $result = $this->db->query(
 162                         "SELECT * FROM
 163                                 $page $pageindex,
 164                                 $revision $revindex,
 165                                 $text
 166                                 WHERE $where $join AND rev_text_id=old_id
 167                                 ORDER BY page_id", $fname );
 168                 $wrapper = $this->db->resultObject( $result );
 169                 $this->outputStream( $wrapper );
 170
 171                 if( $this->buffer == MW_EXPORT_STREAM ) {
 172                         $this->db->bufferResults( $prev );
 173                 }
 174
 175                 wfProfileOut( $fname );
 176         }
 177
 178         /**
 179          * Runs through a query result set dumping page and revision records.
 180          * The result set should be sorted/grouped by page to avoid duplicate
 181          * page records in the output.
 182          *
 183          * The result set will be freed once complete. Should be safe for
 184          * streaming (non-buffered) queries, as long as it was made on a
 185          * separate database connection not managed by LoadBalancer; some
 186          * blob storage types will make queries to pull source data.
 187          *
 188          * @param ResultWrapper $resultset
 189          * @access private
 190          */
 191         function outputStream( $resultset ) {
 192                 $last = null;
 193                 while( $row = $resultset->fetchObject() ) {
 194                         if( is_null( $last ) ||
 195                                 $last->page_namespace != $row->page_namespace ||
 196                                 $last->page_title     != $row->page_title ) {
 197                                 if( isset( $last ) ) {
 198                                         $output = $this->writer->closePage();
 199                                         $this->sink->writeClosePage( $output );
 200                                 }
 201                                 $output = $this->writer->openPage( $row );
 202                                 $this->sink->writeOpenPage( $row, $output );
 203                                 $last = $row;
 204                         }
 205                         $output = $this->writer->writeRevision( $row );
 206                         $this->sink->writeRevision( $row, $output );
 207                 }
 208                 if( isset( $last ) ) {
 209                         $output = $this->writer->closePage();
 210                         $this->sink->writeClosePage( $output );
 211                 }
 212                 $resultset->free();
 213         }
 214 }
 215
 216 class XmlDumpWriter {
 217
 218         /**
 219          * Returns the export schema version.
 220          * @return string
 221          */
 222         function schemaVersion() {
 223                 return "0.3";
 224         }
 225
 226         /**
 227          * Opens the XML output stream's root <mediawiki> element.
 228          * This does not include an xml directive, so is safe to include
 229          * as a subelement in a larger XML stream. Namespace and XML Schema
 230          * references are included.
 231          *
 232          * Output will be encoded in UTF-8.
 233          *
 234          * @return string
 235          */
 236         function openStream() {
 237                 global $wgContLanguageCode;
 238                 $ver = $this->schemaVersion();
 239                 return wfElement( 'mediawiki', array(
 240                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 241                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 242                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 243                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 244                         'version'            => $ver,
 245                         'xml:lang'           => $wgContLanguageCode ),
 246                         null ) .
 247                         "\n" .
 248                         $this->siteInfo();
 249         }
 250
 251         function siteInfo() {
 252                 $info = array(
 253                         $this->sitename(),
 254                         $this->homelink(),
 255                         $this->generator(),
 256                         $this->caseSetting(),
 257                         $this->namespaces() );
 258                 return "  <siteinfo>\n    " .
 259                         implode( "\n    ", $info ) .
 260                         "\n  </siteinfo>\n";
 261         }
 262
 263         function sitename() {
 264                 global $wgSitename;
 265                 return wfElement( 'sitename', array(), $wgSitename );
 266         }
 267
 268         function generator() {
 269                 global $wgVersion;
 270                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 271         }
 272
 273         function homelink() {
 274                 $page = Title::newFromText( wfMsgForContent( 'mainpage' ) );
 275                 return wfElement( 'base', array(), $page->getFullUrl() );
 276         }
 277
 278         function caseSetting() {
 279                 global $wgCapitalLinks;
 280                 // "case-insensitive" option is reserved for future
 281                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 282                 return wfElement( 'case', array(), $sensitivity );
 283         }
 284
 285         function namespaces() {
 286                 global $wgContLang;
 287                 $spaces = "  <namespaces>\n";
 288                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 289                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 290                 }
 291                 $spaces .= "    </namespaces>";
 292                 return $spaces;
 293         }
 294
 295         /**
 296          * Closes the output stream with the closing root element.
 297          * Call when finished dumping things.
 298          */
 299         function closeStream() {
 300                 return "</mediawiki>\n";
 301         }
 302
 303
 304         /**
 305          * Opens a <page> section on the output stream, with data
 306          * from the given database row.
 307          *
 308          * @param object $row
 309          * @return string
 310          * @access private
 311          */
 312         function openPage( $row ) {
 313                 $out = "  <page>\n";
 314                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 315                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 316                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 317                 if( '' != $row->page_restrictions ) {
 318                         $out .= '    ' . wfElement( 'restrictions', array(),
 319                                 strval( $row->page_restrictions ) ) . "\n";
 320                 }
 321                 return $out;
 322         }
 323
 324         /**
 325          * Closes a <page> section on the output stream.
 326          *
 327          * @access private
 328          */
 329         function closePage() {
 330                 return "  </page>\n";
 331         }
 332
 333         /**
 334          * Dumps a <revision> section on the output stream, with
 335          * data filled in from the given database row.
 336          *
 337          * @param object $row
 338          * @return string
 339          * @access private
 340          */
 341         function writeRevision( $row ) {
 342                 $fname = 'WikiExporter::dumpRev';
 343                 wfProfileIn( $fname );
 344
 345                 $out  = "    <revision>\n";
 346                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 347
 348                 $ts = wfTimestamp2ISO8601( strval( $row->rev_timestamp ) );
 349                 $out .= "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 350
 351                 $out .= "      <contributor>\n";
 352                 if( $row->rev_user ) {
 353                         $out .= "        " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n";
 354                         $out .= "        " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n";
 355                 } else {
 356                         $out .= "        " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n";
 357                 }
 358                 $out .= "      </contributor>\n";
 359
 360                 if( $row->rev_minor_edit ) {
 361                         $out .=  "      <minor/>\n";
 362                 }
 363                 if( $row->rev_comment != '' ) {
 364                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 365                 }
 366
 367                 $text = strval( Revision::getRevisionText( $row ) );
 368                 $out .= "      " . wfElementClean( 'text',
 369                         array( 'xml:space' => 'preserve' ),
 370                         strval( $text ) ) . "\n";
 371
 372                 $out .= "    </revision>\n";
 373
 374                 wfProfileOut( $fname );
 375                 return $out;
 376         }
 377
 378 }
 379
 380
 381 /**
 382  * Base class for output stream; prints to stdout or buffer or whereever.
 383  */
 384 class DumpOutput {
 385         function writeOpenStream( $string ) {
 386                 $this->write( $string );
 387         }
 388
 389         function writeCloseStream( $string ) {
 390                 $this->write( $string );
 391         }
 392
 393         function writeOpenPage( $page, $string ) {
 394                 $this->write( $string );
 395         }
 396
 397         function writeClosePage( $string ) {
 398                 $this->write( $string );
 399         }
 400
 401         function writeRevision( $rev, $string ) {
 402                 $this->write( $string );
 403         }
 404
 405         /**
 406          * Override to write to a different stream type.
 407          * @return bool
 408          */
 409         function write( $string ) {
 410                 print $string;
 411         }
 412 }
 413
 414 /**
 415  * Stream outputter to send data to a file.
 416  */
 417 class DumpFileOutput extends DumpOutput {
 418         var $handle;
 419
 420         function DumpFileOutput( $file ) {
 421                 $this->handle = fopen( $file, "wt" );
 422         }
 423
 424         function write( $string ) {
 425                 fputs( $this->handle, $string );
 426         }
 427 }
 428
 429 /**
 430  * Stream outputter to send data to a file via some filter program.
 431  * Even if compression is available in a library, using a separate
 432  * program can allow us to make use of a multi-processor system.
 433  */
 434 class DumpPipeOutput extends DumpFileOutput {
 435         function DumpPipeOutput( $command, $file = null ) {
 436                 if( !is_null( $file ) ) {
 437                         $command .=  " > " . wfEscapeShellArg( $file );
 438                 }
 439                 $this->handle = popen( $command, "w" );
 440         }
 441 }
 442
 443 /**
 444  * Sends dump output via the gzip compressor.
 445  */
 446 class DumpGZipOutput extends DumpPipeOutput {
 447         function DumpGZipOutput( $file ) {
 448                 parent::DumpPipeOutput( "gzip", $file );
 449         }
 450 }
 451
 452 /**
 453  * Sends dump output via the bgzip2 compressor.
 454  */
 455 class DumpBZip2Output extends DumpPipeOutput {
 456         function DumpBZip2Output( $file ) {
 457                 parent::DumpPipeOutput( "bzip2", $file );
 458         }
 459 }
 460
 461 /**
 462  * Sends dump output via the p7zip compressor.
 463  */
 464 class Dump7ZipOutput extends DumpPipeOutput {
 465         function Dump7ZipOutput( $file ) {
 466                 $command = "7za a -si " . wfEscapeShellArg( $file );
 467                 parent::DumpPipeOutput( $command );
 468         }
 469 }
 470
 471
 472
 473 /**
 474  * Dump output filter class.
 475  * This just does output filtering and streaming; XML formatting is done
 476  * higher up, so be careful in what you do.
 477  */
 478 class DumpFilter {
 479         function DumpFilter( &$sink ) {
 480                 $this->sink =& $sink;
 481         }
 482
 483         function writeOpenStream( $string ) {
 484                 $this->sink->writeOpenStream( $string );
 485         }
 486
 487         function writeCloseStream( $string ) {
 488                 $this->sink->writeCloseStream( $string );
 489         }
 490
 491         function writeOpenPage( $page, $string ) {
 492                 $this->sendingThisPage = $this->pass( $page, $string );
 493                 if( $this->sendingThisPage ) {
 494                         $this->sink->writeOpenPage( $page, $string );
 495                 }
 496         }
 497
 498         function writeClosePage( $string ) {
 499                 if( $this->sendingThisPage ) {
 500                         $this->sink->writeClosePage( $string );
 501                         $this->sendingThisPage = false;
 502                 }
 503         }
 504
 505         function writeRevision( $rev, $string ) {
 506                 if( $this->sendingThisPage ) {
 507                         $this->sink->writeRevision( $rev, $string );
 508                 }
 509         }
 510
 511         /**
 512          * Override for page-based filter types.
 513          * @return bool
 514          */
 515         function pass( $page, $string ) {
 516                 return true;
 517         }
 518 }
 519
 520 /**
 521  * Simple dump output filter to exclude all talk pages.
 522  */
 523 class DumpNotalkFilter extends DumpFilter {
 524         function pass( $page ) {
 525                 return !Namespace::isTalk( $page->page_namespace );
 526         }
 527 }
 528
 529 /**
 530  * Dump output filter to include or exclude pages in a given set of namespaces.
 531  */
 532 class DumpNamespaceFilter extends DumpFilter {
 533         var $invert = false;
 534         var $namespaces = array();
 535
 536         function DumpNamespaceFilter( &$sink, $param ) {
 537                 parent::DumpFilter( $sink );
 538
 539                 $constants = array(
 540                         "NS_MAIN"           => NS_MAIN,
 541                         "NS_TALK"           => NS_TALK,
 542                         "NS_USER"           => NS_USER,
 543                         "NS_USER_TALK"      => NS_USER_TALK,
 544                         "NS_PROJECT"        => NS_PROJECT,
 545                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 546                         "NS_IMAGE"          => NS_IMAGE,
 547                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 548                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 549                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 550                         "NS_TEMPLATE"       => NS_TEMPLATE,
 551                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 552                         "NS_HELP"           => NS_HELP,
 553                         "NS_HELP_TALK"      => NS_HELP_TALK,
 554                         "NS_CATEGORY"       => NS_CATEGORY,
 555                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 556
 557                 if( $param{0} == '!' ) {
 558                         $this->invert = true;
 559                         $param = substr( $param, 1 );
 560                 }
 561
 562                 foreach( explode( ',', $param ) as $key ) {
 563                         $key = trim( $key );
 564                         if( isset( $constants[$key] ) ) {
 565                                 $ns = $constants[$key];
 566                                 $this->namespaces[$ns] = true;
 567                         } elseif( is_numeric( $key ) ) {
 568                                 $ns = intval( $key );
 569                                 $this->namespaces[$ns] = true;
 570                         } else {
 571                                 die( "Unrecognized namespace key '$key'\n" );
 572                         }
 573                 }
 574         }
 575
 576         function pass( $page ) {
 577                 $match = isset( $this->namespaces[$page->page_namespace] );
 578                 return $this->invert xor $match;
 579         }
 580 }
 581
 582
 583 /**
 584  * Dump output filter to include only the last revision in each page sequence.
 585  */
 586 class DumpLatestFilter extends DumpFilter {
 587         var $page, $pageString, $rev, $revString;
 588
 589         function writeOpenPage( $page, $string ) {
 590                 $this->page = $page;
 591                 $this->pageString = $string;
 592         }
 593
 594         function writeClosePage( $string ) {
 595                 if( $this->rev ) {
 596                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 597                         $this->sink->writeRevision( $this->rev, $this->revString );
 598                         $this->sink->writeClosePage( $string );
 599                 }
 600                 $this->rev = null;
 601                 $this->revString = null;
 602                 $this->page = null;
 603                 $this->pageString = null;
 604         }
 605
 606         function writeRevision( $rev, $string ) {
 607                 if( $rev->rev_id == $this->page->page_latest ) {
 608                         $this->rev = $rev;
 609                         $this->revString = $string;
 610                 }
 611         }
 612 }
 613
 614 /**
 615  * Base class for output stream; prints to stdout or buffer or whereever.
 616  */
 617 class DumpMultiWriter {
 618         function DumpMultiWriter( $sinks ) {
 619                 $this->sinks = $sinks;
 620                 $this->count = count( $sinks );
 621         }
 622
 623         function writeOpenStream( $string ) {
 624                 for( $i = 0; $i < $this->count; $i++ ) {
 625                         $this->sinks[$i]->writeOpenStream( $string );
 626                 }
 627         }
 628
 629         function writeCloseStream( $string ) {
 630                 for( $i = 0; $i < $this->count; $i++ ) {
 631                         $this->sinks[$i]->writeCloseStream( $string );
 632                 }
 633         }
 634
 635         function writeOpenPage( $page, $string ) {
 636                 for( $i = 0; $i < $this->count; $i++ ) {
 637                         $this->sinks[$i]->writeOpenPage( $page, $string );
 638                 }
 639         }
 640
 641         function writeClosePage( $string ) {
 642                 for( $i = 0; $i < $this->count; $i++ ) {
 643                         $this->sinks[$i]->writeClosePage( $string );
 644                 }
 645         }
 646
 647         function writeRevision( $rev, $string ) {
 648                 for( $i = 0; $i < $this->count; $i++ ) {
 649                         $this->sinks[$i]->writeRevision( $rev, $string );
 650                 }
 651         }
 652 }
 653
 654
 655
 656 function wfTimestamp2ISO8601( $ts ) {
 657         #2003-08-05T18:30:02Z
 658         return preg_replace( '/^(....)(..)(..)(..)(..)(..)$/', '$1-$2-$3T$4:$5:$6Z', wfTimestamp( TS_MW, $ts ) );
 659 }
 660
 661 function xmlsafe( $string ) {
 662         $fname = 'xmlsafe';
 663         wfProfileIn( $fname );
 664
 665         /**
 666          * The page may contain old data which has not been properly normalized.
 667          * Invalid UTF-8 sequences or forbidden control characters will make our
 668          * XML output invalid, so be sure to strip them out.
 669          */
 670         $string = UtfNormal::cleanUp( $string );
 671
 672         $string = htmlspecialchars( $string );
 673         wfProfileOut( $fname );
 674         return $string;
 675 }
 676
 677 ?>